What really counts as a SIFI?

I have been following with great interest the implosion of of Silicon Valley Bank and the threatened contagion to Signature Bank and First Republic. These are very different failures (or almost failures) than what we saw with e.g. Lehman Brothers and the TARP/etc. programs that saved major banks 15 years ago - they're mostly regional banks, mostly sector-heavy with lots of uninsured deposits, and are being exposed to traditional bank runs by flighty deposits as their capital position eroded.

Interestingly, though, SVB at least was done in by having too much exposure to Treasuries, which is kinda hilarious given that Treasuries are generally considered solid and safe assets for tier 1 capital ratios (compared to, say, mortgage backed securities)... but if you don't have enough assets to offset a 550 basis point increase in the fed funds rate, you'll still take a big loss when/if you try to liquidate these assets. Scuttlebutt we heard was that SVB had only stress tested up to a 200 basis point increase, which seems wildly optimistic and probably a good reason for someone to get sued or go to jail (and for some regulators in California to get fired).

What I found interesting was how quickly the feds swooped in and provided a backstop for uninsured depositors (a bailout by any other name) - they were also aggressive in taking over Signature even though it's not clear that Signature had problems anywhere near as severe as SVB. While the size of SVB is pretty big for a failure (only second to WaMu), SFB was not on any list of SIFIs or DFIs, nor would wiping out tens of billions of uninsured deposits (after liquidating the remaining bank assets) necessarily have been all that awful for the economy as a whole.

Of course, SVB is systemically important, mostly because of who was holding their deposits. Something like 50% of the venture backed companies in the US kept most or all of their cash in one bank. If they couldn't access cash to make payroll, they would not have been legally allowed to open their doors on Monday morning. And it's likely that a whole lot of startups would have had to suddenly close up shop. My own company used to have our cash stashed in SVB until last year; then we moved to Signature (heh), though fortunately everything has worked out and we're still a going concern. But you can imagine the carnage that would have happened if over half of the country's startups suddenly went bust.

So it makes a whole lot of sense that the feds made sure the bank would honor deposits and then acted pretty aggressively to quell contagion fears. However, it raises two questions:

1. How can we actually determine what counts as systemically important, with all of the attendant requirements and scrutiny that comes with such a designation? There must be some way to address the moral hazard problem here. SVB might not have been 'too big to fail', but their clients were too important to fail.
2. How can we prevent such risky sectoral concentrations in the future? Our CEO heard a lot of rumors about companies who moved to JPM, widely treated as the safest option out there - we also were directly approached by JPM to move our business with shockingly fast turn times. Our best guess is that JPM saw tens of billions of dollars of deposit inflows in just a few days. But is this really a better solution? Sure, JPM is less likely to experience a sectoral bank run like SVB suffered, and they're much better insulated against changing asset prices (they have a bigger insured deposit base and much better diversified assets and deposits). But that seems like we're just trading one sort of systemic risk for another.

There's a reason why so much venture money is tied up in so few institutions - these banks provide specific services tailored for startups that you don't get at your mom-and-pop regional bank, and that the big behemoths don't bother to develop in any significant degree (since the amounts are still pretty small). It's not obvious how to structure incentives so we avoid this kind of problem in the future.

Some followup: I've been reading a whole gamut of takes on this mess. Senator Warren et al point to the Crapo bill that explicitly stripped some Dodd-Frank oversight from $50-250 billion banks. I think there's some justice to this remark, though it hardly absolved CA regulators from oversight. More generally, I suspect that run of the mill regional banks in this class might in fact be a relatively small systemic risk and could potentially benefit from a laxer regulatory burden... it's really just these weird banks focused on one class of flighty and uninsured deposits that had unique risks. That being said, better risk analysis and stress testing could easily have told SVB not to hold a huge number of long dated securities in an inflationary environment.

I've seen others arguing that the bank should have been allowed to fail because the consequences would not have been all that bad. I think this is probably wrong. Sure, it's likely that most of the deposits would eventually have been returned (my back of the envelope guess is ~90% of uninsured deposits might have eventually been available to depositors), and the scale of that loss is manageable. I think there are two problems with this way of thinking: first, winding up a failed bank (and finding an appropriate buyer for the assets) is not straightforward or particularly fast. As I understand it, a startup that does not have cash on hand to make payroll cannot legally operate even if they theoretically will get cash eventually. One could maybe imagine some bridge loans or other short term funding measures, but getting this scale of fix set up in this time frame may have been infeasible, and a federal backstop might have just solved all of the problems. Secondly, I think contagion is a real risk here - Signature doesn't appear to have been in anywhere near as much trouble as SVB (or FRB for that matter), but I could easily see bank runs starting on large numbers of regional banks even if their deposits and assets are totally differently structured. Dramatic moves by the feds might have been necessary to reassure markets.

That being said, it's still a huge moral hazard problem. Even if bondholder and shareholders are completely hosed and management is appropriately punished, the main lesson other institutions will take away from this is that you can have lax risk management practices because momma will always come to fix things. I do not have a good idea of how to handle this fundamental problem at the heart of our current financial regulatory environment.

I think the more important question is what does this mean about the rate of interest increases. Sure, buying lots of long-dated Treasuries has some risk but it isn't wild or exotic.

It's wild if you only model for a 200 basis point increase. Anyone with half a brain knew the low rates wouldn't stay that way indefinitely.